Fiber-Reinforced Polymers (FRP) have attracted much attention as a promising solution for preserving existing Reinforced Concrete(RC) buildings. Structures could be maintained by reinforcing, repairing, or retrofitting to address seismic inadequacies. For RCbeams, shear failure is identified as the most catastrophic failure mode due to the lack of failure warning. However, there is not enough information on the shear behavior of these retrofitted beams, especially regarding the ideal design and placement of the FRP composites. This study aims to examine the shear strength of RC beams retrofitted with FRP composites and identifies the most efficient design and deployment procedures for these composites. The Artificial Neural Network (ANN) algorithm enhances the precision and efficiency of forecasting the shear strength, increases the solidity and durability of RCstructures, and reduces the need for expensive repairs or replacements. Three RC beams were examined experimentally under combined torsion and shear. ANN values of RMSE = 0.466, R2= 0.856, and r = 0.945 indicate a satisfactory correlation between experimental and numerical values and the AI model's reliability. The results of each training set are near 1 when considering the R2values, regardless of torsion or shear exposure of the retrofitted T-beams. The test set R2values of A1, A2, and AB under torsion and shear demonstrate correct ANN performance. Fiber reinforcement and the volumetric ratio of the FRP materials determine the final structural strength of RC beams enhanced with FRP. Higher torsional reinforced beams have a larger torsional capacity, final angle of twist, and enhanced post-cracking rigidity for a given twist angle.